Question

The exponential smoothing model from part b, with α=0.1, yields a mean squared error of 14.3 for this time series. If we set α equal to 0.05, the mean squared error is 15.9. If we set α equal to 0.3, the mean squared error is 10.5. Which of these three values of α (0.05, 0.1, 0.3) would be the most appropriate for predicting future weekly sales

Answer 1

Since for any model, we check the mean squared error which helps us to understand how good the model is performing for the given dataset.

Mean Square error given by:

Mean Square Error =

• Here n is the number of sample values,
• y is actual output(data we already have) and
• is predicted value by model i.e. predicted by the exponential smoothing model.

The smaller the value of Mean Square Error(MSE) the better the model is.

Now we want to have such α that will help us to minimize the value Mean Square Error(MSE).

So from the above question, we can see that α = 0.3 is having the lowest mean squared error is 10.5. Hence α = 0.3 would be the most appropriate for predicting future weekly sales.